1. Introduction
Under the pronounced increasing trend of the global surface air temperature (SAT), more extreme events have been frequently occurring in winter over China (Zhou et al. 2018; Liu et al. 2021a; Zhou et al. 2021), causing severe epidemiological and economic burdens. In the most recent decade, subseasonal reversals of SAT anomalies (SATA) over China have happened frequently in winter (Wang et al. 2012; Si et al. 2014; Wang et al. 2015; Si et al. 2016; Yan et al. 2022). For example, there was an anomalously cold condition in China before mid-January 2021, but the winter temperature turned to warm after that (Han et al. 2021; Zhang et al. 2021; Dai et al. 2022). Therefore, while challenging, improving the ability to predict SATA over China on the subseasonal scale is essential for both social and economic reasons.
Previous studies have systematically summarized the interannual variability of winter-mean SATA over China and have revealed that the consistent and south–north modes of winter-mean SATA over China are, respectively, related to the East Asian winter monsoon (EAWM) and the Arctic Oscillation (AO) (Zhu et al. 2007; Kang et al. 2009; Xiao et al. 2018). Apart from the impact of planetary-scale atmospheric circulations, winter El Niño–Southern Oscillation (ENSO), North Atlantic sea surface temperature (SST), Arctic sea ice concentration (SIC), and Eurasian snow cover extent (SCE) also modulate the variabilities of winter SATA over China on the interannual time scales (Chen and Sun 2003; Chen et al. 2003; Qu et al. 2006; Zhou and Wang 2008; Chen et al. 2009; Li and Wang 2011; Wang and He 2012; Sun et al. 2014; He et al. 2015; Shi et al. 2015; Zuo et al. 2016; Li et al. 2017; Lu and Zhou 2018; Shi et al. 2021). Based on these vital factors, some statistical and statistical–dynamic prediction models for winter SATA over China have been developed (Ai et al. 2008; Chen et al. 2007; Jia et al. 2010; Dai and Fan 2020). However, the seasonal mean may remove intraseasonal variations. It is necessary to investigate changes and related mechanisms on the subseasonal scale for more accurate climate and weather forecasts.
Various case studies have preliminarily revealed the obvious intermonth variations of winter SATA over China and emphasized the role of subseasonal variations of the EAWM (Wang et al. 2013; Si et al. 2014; Geng et al. 2017; Zhang and Song 2018). The month-to-month variabilities of SATA over China are also associated with other atmospheric circulations (like the AO and the stratospheric polar vortex), which are always affected by the Arctic SIC, ENSO, the Pacific decadal oscillation (PDO), and so on (Si et al. 2014; Zhang and Song 2018). Dai et al. (2019) revealed the combined influences of November Arctic SIC over different regions on the month-to-month variations of winter temperature over Northeast China via tropospheric processes and tropospheric–stratospheric interactions. Furthermore, the relationship between subseasonal winter temperature anomalies over East Asia and the winter Niño-4 phase modulated by the PDO phase has been examined by composite analysis (Xu et al. 2018).
Extended empirical orthogonal function (EEOF) analysis and multivariable empirical orthogonal function analysis have been widely used to study the leading intraseasonal variation modes of early and late winter over East Asia and China (Wei et al. 2014, 2020; Qi and Pan 2021). Wei et al. (2020) indicated the winter SATA over China includes in-phase and out-of-phase evolutions between early and late winter. Propagation directions and energy dispersion of transient and stationary waves contribute to the consistent and reversal mode. Li et al. (2021) further revealed the strengthened relationship between central Pacific ENSO and the reversal mode of the December and January SATA over China after 1997.
Although some research has indicated the characteristics and the related mechanisms of intermonth variations of SATA over China, few studies have focused on its intraseasonal prediction. Yoo et al. (2018) indicated that the intraseasonal prediction skill of winter air temperature in China drops rapidly after one week. Meanwhile, the research on predictions of intermonth inconsistent temperature or warm–cold phase alternations on the subseasonal scale is relatively small compared with predictions of winter temperature on the seasonal and monthly scale over China (Chen et al. 2007; Ai et al. 2008; Jia et al. 2010; Dai and Fan 2020). Liu et al. (2021b) developed three statistical–dynamical prediction models of monthly air temperature over China in winter using the preceding observed autumn SIC over the Arctic and simultaneous SST output by multimodel ensemble results. These models for monthly temperature can improve the prediction skills for winter temperature over China, but inaccuracies still exist due to the different performances of each predicting model for each winter month, especially in years with warm–cold phase alternations. Thus, we focus on different intermonth modes of winter temperature over China to reveal the related factors and physical mechanisms, as well as evaluations of the predictability of the Climate Forecast System, version 2 (CFSv2). Furthermore, prediction models of different intermonth modes of winter temperature over China are developed based on specific mechanisms. In the future, we expect to combine the predictions of winter temperature over China on the seasonal and monthly scale, as well as the consistent and inconsistent intermonth modes, to further improve the skills of the operational forecasts, which is also the motivation and purpose of this study.
The rest of the paper is organized as follows: section 2 introduces the datasets and methods. The characteristics and predictability of the intermonth modes of SATA over China in winter are presented in section 3. The factors and related mechanisms of different modes are illustrated in section 4. Section 5 reports and validates the prediction model results. Conclusions and some further discussion are provided in section 6.
2. Data and methods
a. Data
In this work, a gridded monthly SAT observation dataset with a high horizontal resolution of 0.25° × 0.25° from more than 2400 observation stations in China from the monthly gridded dataset of SAT over China (CN05.1) during the period 1961–2020 is adopted (Wu and Gao 2013). Considering the special terrain effect of the Qinghai–Tibet Plateau and the uncertainty in CN05.1, we remove the grid points with an altitude above 3000 m in this area to discuss the SAT variations in China (Wu and Gao 2013). Atmospheric circulation and mean surface heat flux data are mainly derived from the fifth major global reanalysis produced by ECMWF (ERA5) at the spatial resolution of 1° × 1° from 1979 to the present day (Hersbach et al. 2020). Snow depth and snow albedo data are also from ERA5-Land. The turbulent heat flux is derived from the sum of the mean surface latent heat flux and the sensible heat flux. The turbulent heat flux data are multiplied by −1 to make positive values indicate upward turbulent heat fluxes. In addition, the monthly SIC and SST are from the Met Office Hadley Centre with a horizontal resolution of 1° × 1° (Rayner et al. 2003). The monthly SCE data on an 89 × 89 grid are obtained from Rutgers University (Robinson et al. 2012; Estilow et al. 2015). For the convenience of measurement, the SCE data are interpolated to 1° × 1°.
The CFSv2 from the National Centers for Environmental Prediction is a fully coupled dynamical seasonal prediction system (Saha et al. 2014). Here, the SAT released in November on a Gaussian grid during the period 1982–2019 is used to evaluate the prediction skill for winter SATA over China. To conveniently evaluate the predictability of SATA over China in winter, these data are interpolated to 0.25° × 0.25° like the observations.
In this work, winter is defined as December of the prior year and January–February of the current year. The common period in this work is 1983–2020. All data and indices in this study are detrended before analysis. The climatology of every variable is calculated as its long-term observational average during the period 1991–2020.
b. Methods
1) Diagnostic analysis
EEOF analysis is widely used to reveal the temporal evolution modes of variables (Weare and Nasstrom 1982). In this study, the EEOF method is applied to separate the intermonth evolution modes of winter SATA over China and their corresponding extended principal components (EPCs). We select the first three leading EEOF modes that are statistically distinct from the remaining modes according to the criteria of North et al. (1982) to discuss the intermonth variations.
Linear regression is used to further reveal the causes of the intermonth modes. The horizontal and vertical wave activity flux (WAF) is diagnosed according to Takaya and Nakamura (2001) and Plumb (1985), respectively. The zonal stationary wavenumber is determined according to Jia et al. (2019). Moreover, the Northern Annular Mode (NAM) index is defined as the normalized time coefficient of the first EOF mode of the daily zonal-mean geopotential height anomaly over 0°–90°N (Baldwin and Thompson 2009). In this study, the geopotential heights and temperatures with a horizontal resolution of 1.25° × 1.25° from ERA5 are used to calculate the vertical wave activity flux and NAM index.
2) EEOF-based prediction models
Based on preceding predictors for each mode, statistical prediction models of the first three leading EPCs of winter SATA over China are individually developed via multivariable linear regression. Thus, the individual intermonth variation modes can be predicted. In addition, spatial modes of EEOF1–3 obtained from the observational and predicted EPCs are used to reconstruct the prediction of monthly SATA over China.
To examine the prediction skills of these models, the cross validation with the one-year-out method is also used in this work (Michaelsen 1987). To assess the downscaling models established in this study, the anomaly correlation coefficient (ACC), temporal correlation coefficient (TCC), and ratio of the same sign (RSS) are obtained between the observed and predicted results. The statistical significance is calculated according to the Student’s t test.
3. Intermonth variabilities of SATA over China and their predictability
a. Features of SATA over China
The leading modes of monthly temperature in winter over China exhibit homogeneous variations (abbreviated as SATA−−−/+++), which can, respectively, explain 55.4%, 52.0%, and 63.3% of the total variance in December, January, and February (Figs. 1a–c). However, the regional-mean SATA over China in each winter month characterizes distinct intermonth changes. The consistent variations of SATA over China only have 15 winters (orange dots), accounting for only 39.5% of the total 38 years. The remaining 23 years demonstrate inconsistent variations on intermonth scales (blue dots, characterized as SATA−++/+−−, SATA+−+/−+−, and SATA−−+/++−), accounting for 26.3%, 18.4%, and 15.8% of the total 38 years, respectively (Fig. 1d, Table 1).
(a)–(c) The first spatial mode of the EOF of SATA over China in December, January and February during the period 1983–2020, respectively. The variance explained by the first EOF mode is shown above the top-right-hand corner of each panel. (d) Monthly regional-mean SATA (units: °C) over China in December (blue bars), January (yellow bars), and February (green bars) during 1983–2020. The blue and orange dots are inconsistent and consistent years.
Citation: Journal of Climate 36, 12; 10.1175/JCLI-D-22-0685.1
Temporally inconsistent and consistent years of regional-mean temperature over China during winter months and their prediction results. Years in boldface indicate accurate prediction years for the reconstructed scheme; years in italics indicate accurate prediction years for CFSv2 outputs.
The EEOF is employed for further confirmation of the spatiotemporal distributions of intermonth winter temperature over China (Fig. 2). The first three leading modes, which are selected for further investigation, can explain 32.3%, 16.9%, and 12.5% of the total variance of the SATA variability over China in winter, respectively. EEOF1 demonstrates consistent changes in SATA over China during the winter (Fig. 2a), whereas EEOF2 displays the reversal pattern between December and the following January–February over most regions of China, which may largely reflect the SATA−++/+−− type (Fig. 2b). The temporally consistent variations during winter over Northeast China in EEOF2 may be related to the north–south dipole pattern of SATA, which is one of the leading modes of winter-mean temperature over China (Xiao et al. 2018). The EEOF3 inconsistent mode is both temporally and spatially inconsistent (Fig. 2c). The cold anomalies in December–January and warm anomalies in February orient from northwestern to eastern Inner Mongolia, like the SATA−−+/++− type. The warm anomalies in December and February and cold anomalies in January dominate south China, like the SATA+−+/−+− type. Therefore, the EEOF modes could describe the intermonth evolutions of winter SATA over China well and benefit the further investigations of related mechanisms.
The first three leading EEOF modes and corresponding time series of SATA over China during 1983–2020. (a)–(c) SATA spatial modes of EEOF1–3, respectively, over China in (left) December, (center) January, and (right) February. (d)–(f) Standardized time series of the EPC1–3, respectively. The variance explained by EEOF1–3 is shown above the top-right-hand corner of each panel.
Citation: Journal of Climate 36, 12; 10.1175/JCLI-D-22-0685.1
b. Prediction skill
To evaluate the predictability of the monthly winter SATA over China in global climate models, the spatial distributions of ACCs of monthly SATA over China between observations and CFSv2 outputs are shown in Figs. 3a–c. The ACCs exceeding the 90% confidence level only occur in central China in December, which means CFSv2 can only predict December temperature over central China well. The TCCs of regional-mean SATA over China in December, January, and February are 0.23, −0.16, and −0.09, respectively (not exceeding the 90% confidence level). The multimodel ensemble (CFSv2 and BCC_CSM1.1 m, version 1.1 of the Beijing Climate Center Climate System Model) predictions also show limited skill for monthly SATA over China in winter (Liu et al. 2021b). Moreover, the intermonth variations of CFSv2 outputs are also much weaker than observations. The predicted total numbers of inconsistent and consistent years are 15 and 23, respectively (Fig. 3d). Only 5 years with consistent changes (1987, 1999, 2013, 2017, and 2020) and 1 year with inconsistent variations (1988) are captured by the CFSv2 outputs (Table 1).
(a)–(c) Spatial patterns of ACCs between the observed SATA and the CFSv2 predictions over China in December, January and February, respectively, during 1983–2020. The TCC of the area-averaged SATA over China between observations and the CFSv2 predictions in each month is shown in the bottom-left-hand corner of each panel. The stippled areas denote the 90% confidence level based on the Student’s t test. (d) Monthly regional-mean SATA (units: °C) over China predicted by CFSv2 in December (blue bars), January (yellow bars), and February (green bars) during 1983–2020. The blue and orange dots are inconsistent and consistent years.
Citation: Journal of Climate 36, 12; 10.1175/JCLI-D-22-0685.1
The spatial patterns of the intermonth variation modes of winter SATA over China are also evaluated. The first EEOF mode can explain 51.3% (larger than the observational EEOF1 mode) of the total variance of the predicted SATA over China, showing a consistent pattern similar to the observations during the whole winter (Figs. 2a and 4a). The corresponding EPC1 characterizes decadal variations although the linear trend has been removed, which is inconsistent with mainly interannual variabilities in the observational results (Figs. 2d and 4d). However, for the other two EEOF modes, the spatial patterns from CFSv2 outputs are different from observations. The predicted EEOF2 demonstrates consistently temporal evolutions in winter featuring a dipole pattern between Northeast China and other regions (Fig. 4b). The CFSv2 outputs also fail to predict the reversal change between December and January–February in the eastern part of China (Figs. 2b and 4b). Furthermore, the CFSv2 outputs are unable to accurately capture the spatially consistent variations observed in EEOF3 over northeastern China in January (Figs. 2c and 4c). The predicted time series of observed EEOF1–3 are obtained by projecting CFSv2 outputs onto the observed EEOF1–3 modes of winter SATA over China. The TCCs between the observations and CFSv2 projection results are 0.00, 0.24, and 0.09, respectively (not exceeding the 90% confidence level).
The CFSv2-predicted first three leading EEOF modes and corresponding time series of SATA over China during 1983–2020. (a)–(c) SATA spatial modes of EEOF1–3, respectively, over China in (left) December, (center) January, and (right) February. (d)–(f) Standardized time series of the EPC1–3, respectively. The variance explained by EEOF1–3 is shown above the top-right-hand corner of each panel.
Citation: Journal of Climate 36, 12; 10.1175/JCLI-D-22-0685.1
Thus, the intermonth variation modes of winter SATA over China cannot be predicted well in the CFSv2 outputs. Finding predictors with pronounced statistical relationships and explicit mechanisms are important to develop well-performing prediction models for SATA over China on a monthly scale and improve the prediction skill of dynamic models.
4. Predictors
Arctic SIC, Eurasian SCE, and Atlantic SST are consistently essential factors in winter temperature over China (Li and Wang 2011; Xie et al. 2014; Shi et al. 2015). Zuo et al. (2016) illustrated the importance of autumn Arctic sea ice in predicting winter temperature in China. The impact of Eurasian snow cover on the EAWM and the related winter temperature over Northeast China have also been revealed (Zhang et al. 2016). In addition, the autumn northern Atlantic SST affects winter temperature in China via the periodic air–sea interactions, which further affects winter geopotential anomalies in the Barents Sea and Ural regions (Shi et al. 2015). Therefore, the preceding SIC, SCE, and SST are considered as alternative predictors for the intermonth prediction of SATA over China in winter. The potential connections are indicated in Fig. 5. The consistent EEOF1 mode and the inconsistent EEOF2 and EEOF3 modes of intermonth winter SATA over China are related to the September SIC over the Barents–Laptev Seas (Fig. 5a), the dipole pattern of November SCE over Eurasia (Fig. 5b), and the November SST over the North Atlantic (Fig. 5c), respectively. As physical mechanisms and significant statistical relationships between predictors and predictands are both essential to the skill of statistical prediction models, the related physical mechanisms are described in the following part of this section.
Regression maps of (a) previous September SIC anomalies onto the normalized EPC1, (b) previous November SCE anomalies onto the normalized EPC2, and (c) previous November SST anomalies (units: °C) onto the normalized EPC3 during 1983–2020. The stippled areas denote the 90% confidence level based on the Student’s t test. (d) Normalized September area-averaged SIC over (75°–85°N, 30°–150°E) as BLSICI during 1982–2019. (e), (f) Normalized November area-averaged [two blue boxes in (b)] SCE over (e) (45°–65°N, 20°–55°E) and (f) (35°–55°N, 90°–140°E) as WESCEI and EASCEI during 1982–2019. (g) Normalized November area-averaged [blue box in (c)] SST as NASSTI over (40°–65°N, 60°W–0°) during 1982–2019.
Citation: Journal of Climate 36, 12; 10.1175/JCLI-D-22-0685.1
a. September SIC over the Barents–Laptev Seas related to EEOF1
For further investigation, the SIC index is defined as the regional-mean SIC over the Barents–Laptev Seas (75°–85°N, 30°–150°E), abbreviated as BLSICI (Fig. 5d). The TCC between EPC1 and BLSICI is −0.51, exceeding the 99% confidence level. Previous research has indicated connections between the sea ice over the Barents Sea and East Asian winter temperature by temperature gradient variations and wave–mean flow interaction (Xie et al. 2014; Chen et al. 2014). Lu et al. (2019) presented the significant influence of September SIC over the Barents and Kara Seas on winter SATA over China via oceanic heat storage and the interactions between SIC and the atmosphere during September–February.
Less-than-normal SIC over the Barents–Laptev Seas in the previous September warms surface ocean and lower-tropospheric atmosphere, leading to a less-than-normal SIC condition over the Barents–Laptev Seas during the whole autumn–winter (Figs. 6a,b; Francis et al. 2009). The persistent upward heat flux is intensified by decreases of SIC over the Barents–Laptev Seas during September–February (Fig. 6c). These interactions between SIC and the atmosphere exist during the following months until February, inducing a persistent dipole pattern of geopotential height anomalies over the Barents Sea and East Asia in the middle troposphere, with the southeastward propagation of WAF from the Arctic to East Asia in winter (Fig. 6d). Thus, the westerlies consistently weaken over East Asia in the upper troposphere (figure not shown) and the Siberian high strengthens (Fig. 6e), favoring cold-air masses invading East Asia. Hence, the air temperature in China is consistently lower than normal during December–February.
TCCs of BLSICI between (a) September and every month during October–February, (b) two adjacent months in September–February during 1983–2020. (c) TCCs between negative BLSICI and regional-mean (75°–85°N, 30°–150°E) upward turbulent heat flux index over Barents–Laptev Seas (BLTHFI) between two adjacent months in September–February during 1983–2020. The dotted line denotes the 90% confidence level based on the Student’s t test. Regression maps of monthly (d) 500-hPa geopotential height anomalies (shaded; units: gpm) and horizontal Takaya and Nakamura (2001) WAF (vectors; units: m2 s−2) in December–February and (e) sea level pressure anomalies (units: hPa) in December–February during 1983–2020 onto the normalized preceding negative September BLSICI. The stippled areas denote the 90% confidence level based on the Student’s t test.
Citation: Journal of Climate 36, 12; 10.1175/JCLI-D-22-0685.1
b. The dipole pattern of November SCE over Eurasia connected with EEOF2
With rapidly decreasing sea ice cover, the autumn snow cover over Eurasia demonstrates an increasing trend in recent years (Liu et al. 2012). Meanwhile, some research has emphasized the importance and prediction value of autumn snow cover to winter temperature over Eurasia and intraseasonal precipitation variations in southern China (Han and Sun 2018; Li et al. 2020). The EEOF2 of SATA over China characterizes both spatiotemporally inconsistent variations during winter, and the temperature variations over Northeast China are different from other regions. Although the SCE anomalies related to EEOF2 exhibit a dipole pattern, the domain over East Asia covers the region of Northeast China (Fig. 5b). To further confirm the influence of SCE over western Europe and East Asia on EEOF2, snow-cover indices over western Europe (WESCEI; 45°–65°N, 20°–55°E) and East Asia (EASCEI; 35°–55°N, 90°–140°E) are defined (Figs. 5e,f).
As Fig. 7a depicts, the SCE over western Europe in November has a significant influence on the December SATA over northwestern and southern China, which reverses in January–February. Meanwhile, the local SCE over East Asia in November has a continuous impact on the consistent variations of winter SATA over Northeast China (Fig. 7b). Moreover, the TCC between WESCEI and EASCEI is −0.19 (not exceeding the 90% confidence level), affirming that the SCEs over these domains are independent of each other. The multiple linear regression of EPC2 on WESCEI and EASCEI during the period 1983–2020 are EPC2 = −0.31 × WECEI + 0.44 × EASCEI; therefore, the relative contributions of WESCEI and EASCEI to EEOF2 are 41.38% and 58.62%, respectively. Hence, we hypothesize that the SCE over the different domains of the dipole pattern could individually affect SATA over China via different mechanisms.
Regression maps of monthly SATA (units: °C) over China in (left to right) December–February onto the normalized (a) negative WESCEI and (b) EASCEI in the preceding November during 1983–2020. The stippled areas denote the 90% confidence level based on the Student’s t test.
Citation: Journal of Climate 36, 12; 10.1175/JCLI-D-22-0685.1
1) November SCE over western Europe
To conveniently explain the mechanisms, negative WESCEI is used in Figs. 8 and 9. Based on the radiative cooling effects (Chen and Sun 2003; Chen et al. 2003), the air temperature is anomalously warmer from western Europe to the Ural region in November–December, with persistently decreasing snow cover over western Europe (Figs. 8a,b). Correspondingly, the zonal temperature difference between western Europe and East Asia may intensify the Ural blocking high in December (Figs. 8b,c). With positive–negative geopotential height anomalies over the Urals and East Asia, the 500-hPa zonal WAF propagates from western Europe to East Asia (Fig. 8c). Meanwhile, the 200-hPa zonal wind over Siberia reduces in December (figure not shown). These circulation anomalies all favor cold-air masses invading China, and hence, a colder pattern may occur in December (Fig. 7a). However, the temperature anomalies over western Europe related to snow-cover variations cannot last in the following January–February (figure not shown). Thus, the formation of the EEOF2 mode of SATA over China in January–February should be considered from other aspects.
Regression maps of monthly (a) SCE anomalies, (b) SATA (units: °C), and (c) 500-hPa geopotential height anomalies (shaded; units: gpm) and horizontal Takaya and Nakamura (2001) WAF (vectors; units: m2 s−2) in (left) November and (right) December onto the normalized preceding negative November WESCEI during 1982/83–2019/20. The stippled areas denote the 90% confidence level based on the Student’s t test.
Citation: Journal of Climate 36, 12; 10.1175/JCLI-D-22-0685.1
Regression maps of (a)–(d) monthly meridionally averaged (45°–65°N) Plumb’s (1985) zonal and vertical WAF anomalies (units: m2 s−2) and (e) daily NAM index in November–February onto the normalized preceding negative November WESCEI during 1983–2020. The shaded areas in (a)–(d) and stippled areas in (e) denote the 90% confidence level based on the Student’s t test.
Citation: Journal of Climate 36, 12; 10.1175/JCLI-D-22-0685.1
A series of studies has indicated the influencing mechanisms of snow cover on tropospheric weather, including not only the radiative cooling effect but also troposphere–stratosphere interactions (Chen et al. 2003; Cohen et al. 2007, 2009). Less-than-normal November snow cover over western Europe leads to stronger upward WAF to the stratosphere over western Europe (Figs. 9a,b). Horizontally, eastward propagation of WAF from western Europe to East Asia also strengthens in November–December (Figs. 9a,b). Hence, the stratospheric polar vortex significantly weakens and there are East Asian WAF anomalies downward to the troposphere in January–February (Figs. 9c,d), causing a negative AO-like pattern from the stratosphere to the troposphere (Fig. 9e).
Zuo et al. (2015) indicated that the relationship between the winter AO and SATA in southern China presents a monthly variation in winter. A weakly positive and significantly negative relationship occurs in December and January–February by modulating the relationship between the AO and the Middle East jet stream, respectively. However, the consistent impact of the AO on temperature anomalies in winter in northern China is caused by geopotential anomalies over Siberia. Therefore, with a negative AO-like phase, the SAT demonstrated higher temperature anomalies over southern China in January–February (Fig. 7a).
2) November SCE over East Asia
The influences of SCE over East Asia are also analyzed. Corresponding to the increases of November SCE over East Asia, higher-than-normal local snow albedo dominates during the whole winter (Fig. 10a), favoring greater snow depth over eastern Siberia (Fig. 10b). In this condition, more cold events occur in Northeast China during winter (Fig. 7b). Thus, the monthly variations in temperature are consistent over Northeast China. Meanwhile, a “positive–negative” structure of the meridional difference in air temperature anomalies occurs over the northern Pacific (Fig. 10c). Kuang et al. (2007) indicated that the seasonal variation of the East Asian subtropical westerly jet is closely related to the inhomogeneous pattern of heating. This can be presented by the meridional difference in air temperature anomalies (MDT, which are defined as the differences in air temperature between the two adjacent latitudes with 1° interval; northward is positive). Therefore, the “strong–weak” westerlies pattern with MDT anomalies extends from the Pacific to East Asia between southern and northern China (Figs. 10c,d), preventing cold surges from invading southern China in January–February (Fig. 7b).
Regression maps of (a) snow albedo anomalies, (b) snow depth anomalies (units: cm of water equivalent), (c) 1000–200-hPa averaged MDT anomalies (units: K grid−1), and (d) 200-hPa zonal wind anomalies (units: m s−1) onto the normalized preceding November EASCEI in (left to right) December–February during 1983–2020. The stippled areas denote the 90% confidence level based on the Student’s t test.
Citation: Journal of Climate 36, 12; 10.1175/JCLI-D-22-0685.1
c. November SST over the North Atlantic connected with EEOF3
Qi and Pan (2021) revealed that the intraseasonal phase inversion variation pattern over East Asia is affected by the Eurasian teleconnection pattern phase reversal through the North Atlantic heat flux variations. Therefore, the North Atlantic SST and the generated heat flux variations are important for East Asian intraseasonal modes. Figure 5c demonstrates the positive relationship between North Atlantic SST in the previous November and EEOF3 modes with intermonth inconsistent changes of winter SATA over China. The TCC of the regional-mean SST (NASSTI; 40°–65°N, 60°W–0°) and EPC3 is 0.38, exceeding the 95% confidence level. Figure 11a indicates that the North Atlantic SST has little influence on the December SATA over the whole of China and significant influences on northwestern and southern China in January and on northeastern China in February.
Regression maps of monthly (a) SATA (units: °C) over China in (left to right) December–February, and zonally averaged (60°W–0°) (b) 1000-hPa zonal wind anomalies (shaded; units: m s−1), (c) SST anomalies (units: °C), and (d) upward turbulent heat flux (units: W m−2) in November–February onto the normalized preceding November NASSTI during 1983–2020. The contours in (b) are the 1000-hPa zonal wind climatology (units: m s−1). The stippled areas denote the 90% confidence level based on the Student’s t test.
Citation: Journal of Climate 36, 12; 10.1175/JCLI-D-22-0685.1
According to the “wind–evaporation–SST” feedback mechanism, the anomalous zonal wind at 1000 hPa over the subtropical North Atlantic contrasts with the zonal wind climatology hindering the surface evaporation, making persistently warm SST anomalies in November–December (Figs. 11b,c). Thus, the turbulent heat flux decreases during November–December, resulting in atmospheric interactions with the ocean (Fig. 11d). While the anomalous zonal wind corresponds to the zonal wind climatology, strong releases of upward turbulent heat flux anomalies favor the interaction from the ocean to the atmosphere, triggering the generation and propagation of Rossby waves in January–February (Figs. 11b–d).
The propagation directions of the horizontal WAF show zonal and meridional tracks in January and February, respectively (Fig. 12a). According to the climatological differences of zonal wind at 200 hPa, the westerlies in January are more robust than in February over the midlatitude North Atlantic (Fig. 12c), providing a better background for the zonal propagation of Rossby waves in January. The difference in zonal stationary wavenumber can be interpreted as a refractive index for Rossby waves and regions, with its maxima acting as waveguides (Hoskins and Ambrizzi 1993; Jia et al. 2019). In Fig. 12d, the positive values can be observed over the higher-latitude North Atlantic (50°–70°N, 70°W–10°E), favoring a northward propagation of the NASSTI-related wave train in February.
Regression maps of monthly (a) 200-hPa geopotential height anomalies (shaded; units: gpm) and horizontal Takaya and Nakamura (2001) WAF (vectors; units: m2 s−2) and (b) 200-hPa zonal wind anomalies (units: m s−1) onto the normalized preceding November NASSTI in (left) January (right) February during 1983–2020. The stippled areas denote the 90% confidence level based on the Student’s t test. Differences in (c) climatological 200-hPa zonal wind (units: m s−1) and (d) climatological 200-hPa zonal stationary wavenumber between February and January.
Citation: Journal of Climate 36, 12; 10.1175/JCLI-D-22-0685.1
Therefore, these variations contribute to the propagations of the zonal WAF to the downstream area in January. Consequently, the negative–positive geopotential anomalies over the North Atlantic–Urals and weak zonal wind around Siberia at 200 hPa could intensify the Ural blocking high and Siberian high, favoring more air masses invading China and causing cold anomalies in January (Figs. 12a,b). In contrast, the anomalous circulations exhibit more meridional propagation in February, with weak westerlies over the midlatitude Atlantic and strong westerlies extending from the higher-latitude Atlantic to Siberia, which hinders cold air masses from further entering China (Fig. 12b). Therefore, the warm conditions occur in February over China, especially over northeastern China (Fig. 11a).
5. Verification of the prediction schemes
a. Prediction results for consistent and inconsistent modes
Based on the specific physical mechanisms proposed above, the September BLSICI, November WESCEI and EASCEI, and November NASSTI are adopted to construct prediction schemes separately for the leading modes of intermonth variations of winter SATA over China, respectively abbreviated as EEOF1 scheme, EEOF2 scheme, and EEOF3 scheme. As EEOF1–3 modes are orthogonal to each other, their prediction schemes are independent and are not affected by predictors from other modes.
To estimate the prediction skills of these schemes, the one-year-out cross-validation method is used. For the consistent mode EEOF1, the TCC of EPC1 between the observations and cross-validation results is 0.48, exceeding the 99% confidence level (Fig. 13a). The RSS of the EEOF1 scheme is 81.6% during the period 1983–2020, which means the scheme can accurately reflect the consistent phase of SATA over China in winter. Although one EEOF mode only partly explains the intermonth variations of winter SATA over China, it may dominate the pattern of SATA variations if the corresponding coefficient is extremely high. Thus, the RSSs of anomalous years (defined as the standardized time series of the EPC1–3 below −0.5 standard deviations or higher +0.5 standard deviations) are also calculated. The EEOF1 scheme can also accurately predict 46.4% of anomalous years for the consistent mode EEOF1 (Table 2).
(a)–(c) The standardized cross validations of the EEOF1, EEOF2, and EEOF3 schemes (blue bars) and their corresponding observational results (red bars) for EPC1–3 of SATA over China during 1983–2020. The TCC between the cross-validation and observation results by the EPC1–3 is shown above the top-right-hand corner of each panel, with one asterisk (*) and three asterisks (***) denoting statistical significance at the 90% and 99% confidence level, respectively, based on the Student’s t test.
Citation: Journal of Climate 36, 12; 10.1175/JCLI-D-22-0685.1
RSSs of the cross-validation results of inconsistent and consistent modes.
For EPC2 and EPC3, the TCCs between the observation and cross-validation results are 0.51 and 0.31, exceeding the 99% and 90% confidence levels, respectively (Figs. 13b,c). The RSS of all years is 68.4% and 63.2% for the EEOF2 and EEOF3 scheme, respectively, which are lower than the consistent mode. However, the EEOF2 and EEOF3 schemes outperform the EEOF1 scheme in predicting the anomalous years. The RSSs of the anomalous years are 54.2% (EEOF2 scheme) and 47.6% (EEOF3 scheme). Consequently, inconsistent and consistent modes can be well predicted based on their related predictors.
b. Prediction results for monthly SATA over China and specific intermonth evolution years
Chu et al. (2008) used the respective EOF modes and principal components of their fields to reconstruct the large-scale variables and regional temperature. Dai and Fan (2020) revealed that the reconstructed field, retaining the leading modes that account for at least 90% of the explained variance, can efficiently develop a prediction model for winter temperature in China. As the first three leading EEOF modes together explain 61.7% of the intermonth variance in winter SATA over China, a reconstructed scheme is preliminarily developed using observed EEOF patterns and predicting EPCs.
The TCCs and ACCs are also used to evaluate the prediction skill of the reconstructed scheme. The TCCs of regional-mean SATA over China between the observation and cross-validation results of the reconstructed scheme in December, January, and February are 0.40, 0.27, and 0.45, respectively (all exceeding the 90% confidence level), which efficiently improves the monthly prediction skill for winter SATA over China compared with that of CFSv2 outputs (0.23, −0.16, and −0.09, respectively, not exceeding the 90% confidence level). Moreover, this scheme can significantly improve the prediction skill for SATA in southern China and northeastern China in December (Fig. 14a). In January, western China and northeastern China can be better predicted (Fig. 14b). In February, the SATA can well predict the whole of China (Fig. 14c). Moreover, the winter-mean SATA over China can be obtained from monthly prediction results. The TCC of regional-mean SATA over China between the observation and cross-validation results of the reconstructed field scheme is 0.59, which is higher than that in December, January, and February (0.40, 0.27, and 0.45, respectively). Therefore, the reconstructed field scheme can efficiently promote the prediction skill for monthly and seasonal SATA over China in winter.
(a)–(c) Spatial patterns of ACCs between the observed SATA over China and the cross-validation results of the reconstructed scheme in December, January and February, respectively, during 1983–2020. The stippled areas denote the 90% confidence level based on the Student’s t test. The TCCs of area-averaged SATA over China between the observation and cross-validation results in each month are shown in the bottom-left-hand corner of each panel, with one asterisk (*), two asterisks (**), and three asterisks (***) denoting statistical significance at the 90%, 95%, and 99% confidence level, respectively, based on the Student’s t test. (d) Monthly regional-mean SATA (units: °C) over China hindcasted by the reconstructed scheme in December (blue bars), January (yellow bars), and February (green bars) during 1983–2020. The blue and orange dots are inconsistent and consistent years.
Citation: Journal of Climate 36, 12; 10.1175/JCLI-D-22-0685.1
The number of years with different types of intermonth SATA variations in winter over China is also assessed (Table 1 and Fig. 14d). For inconsistent years, the reconstructed scheme can correctly predict 6 years, which is higher than the 1 year of CFSv2 outputs (Table 1). In particular, the reconstructed scheme can correctly hindcast 3 (out of 10 in the observations) and 2 (out of 7) years for types SATA−++/+−− and SATA+−+/−+−, whereas CFSv2 fails to capture these intraseasonal variation patterns. Although CFSv2 predicts 5 years for consistent variations, 4 years belong to consistently warming years. It has limitations in predicting consistent cooling years. Notably, the reconstructed scheme can promote the prediction of consistently cold years from 1 year (CFSv2 results) to 5 years (reconstructed scheme). Therefore, the reconstructed scheme can efficiently improve the prediction skill for both the temporal and spatial patterns of intermonth variations of winter SATA over China.
c. Case study
Specific years are selected for further assessment of the reconstructed scheme. In the observation, the type SATA−++/+−− is the most common among all inconsistent types of winter SATA over China. Thus, we select 2008, one of the SATA−++/+−− years, to further verify the prediction skill for inconsistent variation years.
The SATA over China in December is warmer than normal, but it turns to cold anomalies in January–February in winter 2008 (Fig. 15a). The CFSv2 prediction results show consistent cold anomalies in winter, except for warm anomalies in a small area of southern China in December (Fig. 15b). The reconstructed scheme can capture the intraseasonal variations (Fig. 15c). It can demonstrate positive anomalies in December and negative anomalies in January and February. Meanwhile, the normalized WESCEI, EASCEI, and NASSTI are 1.34, −1.93, and 2.28, respectively, which are attributed to the intermonth difference of SATA over China in 2008. Thus, the SATA over China in 2008 presents warm anomalies in December and cold anomalies in January and February. However, the temperatures and their anomalies are higher (lower) than the observations over northwestern (northeastern) China in December and is higher than the observations over eastern China in January–February (Fig. 15c). This is due to the first three leading EEOF modes of winter SATA over China only accounting for 61.7% of the intermonth variance. The reconstructed scheme may more accurately reflect the intermonth inconsistent variation patterns.
Spatial patterns of SATA (units: °C) over China in (left) December 2007 and (center) January and (right) February 2008 from (a) observations, (b) CFSv2 outputs, and (c) cross-validation results based on the reconstructed scheme. The contours in (c) are differences of SATA (units: °C) over China between cross-validation results based on the reconstructed scheme and observations in December 2007 and January and February 2008.
Citation: Journal of Climate 36, 12; 10.1175/JCLI-D-22-0685.1
6. Conclusions and discussion
Recently, subseasonal variations and their causes have received increasing attention. However, their predictability remains a big challenge. In this study, the features of the inconsistent and consistent modes of intermonth SATA over China in winter are investigated by the EEOF method. Thus, one consistent mode (EEOF1) and two inconsistent modes (EEOF2 and EEOF3) of winter SATA over China are captured (Fig. 2). Prediction models based on specific physical mechanisms for separate EEOF modes and the reconstructed field are also developed, and exhibit better forecasting skill than CFSv2 outputs.
As in the analyses mentioned above, the consistent mode, EEOF1, is significantly affected by the previous September SIC over the Barents–Laptev Seas via wave–mean flow interactions and air temperature gradient changes. Two inconsistent modes, EEOF2 and EEOF3, have significant correlations with November SCE and SST, respectively. EEOF2 is related to the combined influences of SCE over western Europe and East Asia. November SCE over western Europe impacts SATA over China via the propagations of tropospheric Rossby waves in December and stratospheric–tropospheric interactions in January–February. The East Asian SCE in November causes local radiation cooling effects on winter SATA over China, changing the westerlies’ intensity, which further impacts SATA over southern China in January–February. The North Atlantic SST in November connects with EEOF3 via propagation directions of Rossby waves and zonal circulation variations.
Thus, the September BLSICI, November WESCEI and EASCEI, and November NASSTI are used to develop prediction models for the consistent mode (EEOF1 scheme) and inconsistent modes (EEOF2 and EEOF3 schemes), separately. The validation results show that the TCCs of the time series between the observations and three schemes are 0.48 (EEOF1 scheme), 0.51 (EEOF2 scheme), and 0.31 (EEOF3 scheme) (exceeding the 90% confidence level), and the RSSs in all years are 81.6%, 68.4%, and 63.2%, respectively. Therefore, schemes based on individual predictors for separate modes of winter SATA over China exhibit good performance for both consistent and inconsistent variations. Moreover, monthly air temperature predictions over China in winter are also reconstructed by the observed patterns and corresponding predicted EPCs (reconstructed scheme). The TCCs of SATA over China in December, January, and February between the observation and reconstructed scheme are 0.40, 0.27, and 0.45, respectively (exceeding the 90% confidence level), which are much higher than that of the CFSv2 outputs. Moreover, the reconstructed scheme can accurately predict 6 years with intraseasonal variations, which is more than the CFSv2 outputs can (1 year). Therefore, the prediction models proposed in this study all outperform the CFSv2 outputs in forecasting different patterns of intermonth variations of winter SATA over China, as well as the original SATA on the monthly scale.
However, these three leading EEOF modes of winter SATA over China only account for 61.7% of the intermonth variance, causing a weaker interannual variability of the predicted monthly SATA than observed over China. Some case studies showed that the reversal of SATA over China from a warm (cold) to cold (warm) pattern occurred in January, as in 2015 and 2021 (Wang et al. 2015; Dai and Fan 2022; Yan et al. 2022), which causes predictive difficulties of spatial EEOF modes in January. Thus, the reconstructed scheme has a better prediction skill for monthly SATA over China in December and February than in January. Moreover, this research only takes intermonth variation modes into consideration, meaning the reconstruction results mostly reflect temporal intermonth variation anomalies. Thus, the remaining variance of SATA over China should be further considered in terms of spatial distribution in future work. In addition, some nonlinear effects of intraseasonal SATA over China should be further explored. Meanwhile, we will further evaluate other climate models (e.g., BCC_CSM; GloSea5, Met Office Global Seasonal forecasting system version 5) for intermonth variations of SATA over China in winter to find other effective simultaneous predictors. Based on our understanding of the intermonth modes of winter SATA over China and related factors, the adoption of certain machine learning methods may be useful for further improving the subseasonal prediction skill for winter air temperature in China.
Acknowledgments.
This study was supported by the National Key Research and Development Program of China (Grant 2022YFE0106800), the National Natural Science Foundation of China (NSFC) (Grants 42088101 and 41730964), the Shanghai Sailing Program (Grant 21YF1452000), and the Innovation Group Project of the Southern Marine Science and Engineering Guangdong Laboratory (Zhuhai) (Grant 311021001).
Data availability statement.
The monthly gridded dataset of SAT over China (CN05.1) is provided by the China Meteorological Administration. The ERA5 dataset is available at https://cds.climate.copernicus.eu/cdsapp#!/search?text=ERA5%20back%20extension&type=dataset. The monthly SST and SIC datasets are provided by the Met Office Hadley Centre and can be downloaded from their website at https://www.metoffice.gov.uk/hadobs/hadisst/. The monthly SCE from Rutgers University is obtained at https://climate.rutgers.edu/snowcover/docs.php?target=datareq. The CFSv2 data were derived from http://iridl.ldeo.columbia.edu/SOURCES/.NOAA/.NCEP/.EMC/.CFSv2/.
REFERENCES
Ai, W., W. Dong, and P. Zhang, 2008: A downscaling method based on the empirical orthogonal function (EOF) and its application in seasonal prediction (in Chinese). J. Trop. Meteor., 24, 320–326, https://doi.org/10.16032/j.issn.1004-4965.2008.04.002.
Baldwin, M. P., and D. W. J. Thompson, 2009: A critical comparison of stratosphere–troposphere coupling indices. Quart. J. Roy. Meteor. Soc., 135, 1661–1672, https://doi.org/10.1002/qj.479.
Chen, H., and Z. Sun, 2003: The effects of Eurasian snow cover anomaly on winter atmospheric general circulation. Part I. Observational studies (in Chinese). Chin. J. Atmos. Sci., 27, 304–316, https://doi.org/10.3878/j.issn.1006-9895.2003.03.02.
Chen, H., Z. Sun, and W. Zhu, 2003: The effects of Eurasian snow cover anomaly on winter atmospheric general circulation. Part II. Model simulation (in Chinese). Chin. J. Atmos. Sci., 27, 847–860, https://doi.org/10.3878/j.issn.1006-9895.2003.05.06.
Chen, S. Y., Y. Zhang, Q. Xia, D. Y. Bai, and X. F. Zhang, 2009: Analysis of relationship between winter air temperature in eastern China and sea surface temperature anomaly (in Chinese). Plateau Meteor., 28, 1181–1188.
Chen, X. L., H. B. Wu, G. L. Ding, and X. He, 2007: Ensemble canonical correlation prediction method of winter temperature over China (in Chinese). Trans. Atmos. Sci., 30, 623–631.
Chen, Z., R. Wu, and W. Chen, 2014: Impacts of autumn Arctic sea ice concentration changes on the East Asian winter monsoon variability. J. Climate, 27, 5433–5450, https://doi.org/10.1175/JCLI-D-13-00731.1.
Chu, J.-L., H. Kang, C.-Y. Tam, C.-K. Park, and C.-T. Chen, 2008: Seasonal forecast for local precipitation over northern Taiwan using statistical downscaling. J. Geophys. Res., 113, D12118, https://doi.org/10.1029/2007JD009424.
Cohen, J., M. Barlow, P. J. Kushner, and K. Saito, 2007: Stratosphere–troposphere coupling and links with Eurasian land surface variability. J. Climate, 20, 5335–5343, https://doi.org/10.1175/2007JCLI1725.1.
Cohen, J., M. Barlow, and K. Saito, 2009: Decadal fluctuations in planetary wave forcing modulate global warming in late boreal winter. J. Climate, 22, 4418–4426, https://doi.org/10.1175/2009JCLI2931.1.
Dai, G., C. Li, Z. Han, D. Luo, and Y. Yao, 2022: The nature and predictability of the East Asian extreme cold events of 2020/21. Adv. Atmos. Sci., 39, 566–575, https://doi.org/10.1007/s00376-021-1057-3.
Dai, H., and K. Fan, 2020: Skilful two-month-leading hybrid climate prediction for winter temperature over China. Int. J. Climatol., 40, 4922–4943, https://doi.org/10.1002/joc.6497.
Dai, H., and K. Fan, 2022: Subseasonal reversal of winter temperature over northeast China in 2014/2015: Role of Arctic sea ice. Front. Environ. Sci., 10, 852673, https://doi.org/10.3389/fenvs.2022.852673.
Dai, H., K. Fan, and J. Liu, 2019: Month-to-month variability of winter temperature over Northeast China linked to sea ice over the Davis Strait–Baffin Bay and the Barents–Kara Sea. J. Climate, 32, 6365–6384, https://doi.org/10.1175/JCLI-D-18-0804.1.
Estilow, T. W., A. H. Young, and D. A. Robinson, 2015: A long-term Northern Hemisphere snow cover extent data record for climate studies and monitoring. Earth Syst. Sci. Data, 7, 137–142, https://doi.org/10.5194/essd-7-137-2015.
Francis, J. A., W. Chan, D. J. Leathers, J. R. Miller, and D. E. Veron, 2009: Winter Northern Hemisphere weather patterns remember summer Arctic sea-ice extent. Geophys. Res. Lett., 36, L07503, https://doi.org/10.1029/2009GL037274.
Geng, X., W. Zhang, M. F. Stuecker, and F.-F. Jin, 2017: Strong sub-seasonal wintertime cooling over East Asia and Northern Europe associated with super El Niño events. Sci. Rep., 7, 3770, https://doi.org/10.1038/s41598-017-03977-2.
Han, R., L. Shi, and Y. Yuan, 2021: Analysis on the causes of cold and warm transition in China during the winter of 2020/2021 (in Chinese). Meteor. Mon., 47, 880–892, https://doi.org/10.7519/j.issn.1000-0526.2021.07.011.
Han, S., and J. Sun, 2018: Impacts of autumnal Eurasian snow cover on predominant modes of boreal winter surface air temperature over Eurasia. J. Geophys. Res. Atmos., 123, 10 076–10 091, https://doi.org/10.1029/2018JD028443.
He, J.-H., F.-M. Wu, and L. Qi, 2015: Decadal/interannual linking between autumn Arctic sea ice and following winter Eurasian air temperature (in Chinese). Chin. J. Geophys., 58, 1089–1102, https://doi.org/10.6038/cjg20150401.
Hersbach, H., and Coauthors, 2020: The ERA5 global reanalysis. Quart. J. Roy. Meteor. Soc., 146, 1999–2049, https://doi.org/10.1002/qj.3803.
Hoskins, B. J., and T. Ambrizzi, 1993: Rossby wave propagation on a realistic longitudinally varying flow. J. Atmos. Sci., 50, 1661–1671, https://doi.org/10.1175/1520-0469(1993)050<1661:RWPOAR>2.0.CO;2.
Jia, X., L. Chen, W. Li, and D. Chen, 2010: Statistical downscaling based on BP-CCA: Predictability and application to the winter temperature and precipitation in China (in Chinese). Acta Meteor. Sin., 68, 398–410, https://doi.org/10.11676/qxxb2010.039.
Jia, X., Y. You, R. Wu, and Y. Yang, 2019: Interdecadal changes in the dominant modes of the interannual variation of spring precipitation over China in the mid-1980s. J. Geophys. Res. Atmos., 124, 10 676–10 695, https://doi.org/10.1029/2019JD030901.
Kang, L., W. Chen, L. Wang, and L. Chen, 2009: Interannual variations of winter temperature in China and their relationship with the atmospheric circulation and sea surface temperature (in Chinese). Climatic Environ. Res., 14, 45–53, https://doi.org/10.3878/j.issn.1006-9585.2009.01.05.
Kuang, X., Y. Zhang, and J. Liu, 2007: Seasonal variations of the East Asian subtropical westerly jet and the thermal mechanism. J. Meteor. Res., 21, 192–203.
Li, D., and C. Wang, 2011: Research progress of snow cover and its influence on China climate (in Chinese). Trans. Atmos. Sci., 34, 627–636, https://doi.org/10.13878/j.cnki.dqkxxb.2011.05.013.
Li, H., H. Wang, and D. Jiang, 2017: Influence of October Eurasian snow on winter temperature over Northeast China. Adv. Atmos. Sci., 34, 116–126, https://doi.org/10.1007/s00376-016-5274-0.
Li, H., K. Fan, S. He, Y. Liu, X. Yuan, and H. Wang, 2021: Intensified impacts of central Pacific ENSO on the reversal of December and January surface air temperature anomaly over China since 1997. J. Climate, 34, 1601–1618, https://doi.org/10.1175/JCLI-D-20-0048.1.
Li, J., F. Li, and H. Wang, 2020: Subseasonal prediction of winter precipitation in southern China using the early November snowpack over the Urals. Atmos. Oceanic Sci. Lett., 13, 534–541, https://doi.org/10.1080/16742834.2020.1824547.
Liu, J., J. A. Curry, H. Wang, M. Song, and R. M. Horton, 2012: Impact of declining Arctic sea ice on winter snowfall. Proc. Natl. Acad. Sci. USA, 109, 4074–4079, https://doi.org/10.1073/pnas.1114910109.
Liu, M., Y. Li, and C. Lu, 2021a: Analysis of the characteristics and atmospheric circulation causes of two types of extreme cold events in winter in China (in Chinese). Plateau Meteor., 40, 603–620, https://doi.org/10.7522/j.issn.1000-0534.2020.00093.
Liu, Y., K. Fan, L. Chen, H.-L. Ren, Y. Wu, and C. Liu, 2021b: An operational statistical downscaling prediction model of the winter monthly temperature over China based on a multi-model ensemble. Atmos. Res., 249, 105262, https://doi.org/10.1016/j.atmosres.2020.105262.
Lu, C., and B. Zhou, 2018: Influences of the 11-yr sunspot cycle and polar vortex oscillation on observed winter temperature variations in China. J. Meteor. Res., 32, 367–379, https://doi.org/10.1007/s13351-018-7101-2.
Lu, C., K. Li, S. Xie, Z. Wang, and Y. Qin, 2019: Month-to-month variability of autumn sea ice in the Barents and Kara Seas and its relationship to winter air temperature in China. Adv. Meteor., 2019, 4381438, https://doi.org/10.1155/2019/4381438.
Michaelsen, J., 1987: Cross-validation in statistical climate forecast models. J. Climate Appl. Meteor., 26, 1589–1600, https://doi.org/10.1175/1520-0450(1987)026<1589:CVISCF>2.0.CO;2.
North, G. R., T. L. Bell, R. F. Cahalan, and F. J. Moeng, 1982: Sampling errors in the estimation of empirical orthogonal functions. Mon. Wea. Rev., 110, 699–706, https://doi.org/10.1175/1520-0493(1982)110<0699:SEITEO>2.0.CO;2.
Plumb, R. A., 1985: On the three-dimensional propagation of stationary waves. J. Atmos. Sci., 42, 217–229, https://doi.org/10.1175/1520-0469(1985)042<0217:OTTDPO>2.0.CO;2.
Qi, L., and W. Pan, 2021: Variability of the phase reversal of the East Asia temperature from early to late winter and the possible influencing factors (in Chinese). Chin. J. Atmos. Sci., 45, 1039–1056, https://doi.org/10.3878/j.issn.1006-9895.2011.20181.
Qu, J., Z. Jiang, G. Tan, and L. Sun, 2006: Relation between interannual, interdecadal variability of SST in North Atlantic in winter and air temperature in China (in Chinese). Sci. Geogr. Sin., 26, 557–563.
Rayner, N. A., D. E. Parker, E. B. Horton, C. K. Folland, L. V. Alexander, D. P. Rowell, E. C. Kent, and A. Kaplan, 2003: Global analyses of sea surface temperature, sea ice, and night marine air temperature since the late nineteenth century. J. Geophys. Res., 108, 4407, https://doi.org/10.1029/2002JD002670.
Robinson, D. A., T. W. Estilow, and NOAA CDR Program, 2012: NOAA climate data record (CDR) of Northern Hemisphere (NH) snow cover extent (SCE), version 1. NOAA National Centers for Environmental Information, accessed 9 December 2020, https://doi.org/10.7289/V5N014G9.
Saha, S., and Coauthors, 2014: The NCEP Climate Forecast System version 2. J. Climate, 27, 2185–2208, https://doi.org/10.1175/JCLI-D-12-00823.1.
Shi, C., W. Sun, and D. Guo, 2021: Synergistic effects of WP and NAO on winter surface temperature in southeastern China (in Chinese). Trans. Atmos. Sci., 44, 394–404, https://doi.org/10.13878/j.cnki.dqkxxb.2019122700.
Shi, X., J. Sun, Y. Sun, W. Bi, X. Zhou, and L. Yi, 2015: Impact of the autumn Atlantic sea surface temperature three-pole structure on winter atmospheric circulation (in Chinese). Acta Oceanol. Sin., 37, 33–40, https://doi.org/10.3969/j.issn.0253-4193.2015.07.004.
Si, D., Q. Li, Y. Liu, Z. Wang, Y. Yuan, and D. Wang, 2014: Possible causes for the anomalous weak East Asian winter monsoon in 2013/2014 (in Chinese). Meteor. Mon., 40, 891–897, https://doi.org/10.7519/j.issn.1000-0526.2014.07.014.
Si, D., L. Ma, P. Wang, Y. Wang, Y. Nie, and L. Sun, 2016: Anomalous activity of Arctic Oscillation in winter 2015/2016 and its impact on temperature in China (in Chinese). Meteor. Mon., 42, 892–897.
Sun, S., G. Liu, W. Song, and J. Peng, 2014: A precursory signal for the dipole mode of winter temperature anomaly over eastern China (in Chinese). Chin. J. Atmos. Sci., 38, 727–741, https://doi.org/10.3878/j.issn.1006-9895.2013.13211.
Takaya, K., and H. Nakamura, 2001: A formulation of a phase-independent wave-activity flux for stationary and migratory quasigeostrophic eddies on a zonally varying basic flow. J. Atmos. Sci., 58, 608–627, https://doi.org/10.1175/1520-0469(2001)058<0608:AFOAPI>2.0.CO;2.
Wang, D., B. Zhou, C. Sun, Y. Yuan, Y. Liu, and P. Wang, 2013: Features and possible causes for East Asian winter monsoon in 2012/2013 (in Chinese). Meteor. Mon., 39, 930–937, https://doi.org/10.7519/j.issn.1000-0526.2013.07.014.
Wang, D., T. Cui, D. Si, X. Shao, Q. Li, and C. Sun, 2015: Features and possible causes for East Asian winter monsoon in 2014/2015 (in Chinese). Meteor. Mon., 41, 907–914, https://doi.org/10.7519/j.issn.1000-0526.2015.07.013.
Wang, H., and S. He, 2012: Weakening relationship between East Asian winter monsoon and ENSO after mid-1970s. Chin. Sci. Bull., 57, 3535–3540, https://doi.org/10.1007/s11434-012-5285-x.
Wang, X., Z. Sun, B. Hu, Y. Tan, and G. Zeng, 2012: Relationship between Arctic sea ice thickness distribution and climate of China. Acta Meteor. Sin., 26, 189–204, https://doi.org/10.1007/s13351-012-0205-1.
Weare, B. C., and J. S. Nasstrom, 1982: Examples of extended empirical orthogonal function analyses. Mon. Wea. Rev., 110, 481–485, https://doi.org/10.1175/1520-0493(1982)110<0481:EOEEOF>2.0.CO;2.
Wei, W., L. Wang, Q. Chen, and Y. Liu, 2014: Interannual variations of early and late winter temperatures in China and their linkage (in Chinese). Chin. J. Atmos. Sci., 38, 524–536, https://doi.org/10.3878/j.issn.1006-9895.1401.13320.
Wei, W., L. Wang, Q. Chen, Y. Liu, and Z. Li, 2020: Definition of early and late winter and associated interannual variations of surface air temperature in China (in Chinese). Chin. J. Atmos. Sci., 44, 122–137, https://doi.org/10.3878/j.issn.1006-9895.1904.18238.
Wu, J., and X.-J. Gao, 2013: A gridded daily observation dataset over China region and comparison with the other datasets (in Chinese). Chin. J. Geophys., 56, 1102–1111, https://doi.org/10.6038/cjg20130406.
Xiao, D., Z. Zuo, R. Zhang, X. Zhang, and Q. He, 2018: Year-to-year variability of surface air temperature over China in winter. Int. J. Climatol., 38, 1692–1705, https://doi.org/10.1002/joc.5289.
Xie, Y., Y. Liu, and J. Huang, 2014: The influence of the autumn Arctic sea ice on winter air temperature in China (in Chinese). Acta Meteor. Sin., 72, 703–710, https://doi.org/10.11676/qxxb2014.057.
Xu, X., F. Li, S. He, and H. Wang, 2018: Subseasonal reversal of East Asian surface temperature variability in winter 2014/15. Adv. Atmos. Sci., 35, 737–752, https://doi.org/10.1007/s00376-017-7059-5.
Yan, H., Y. Yuan, G. Tan, and Y. Zi, 2022: Possible impact of sudden stratospheric warming on the intraseasonal reversal of the temperature over East Asia in winter 2020/21. Atmos. Res., 268, 106016, https://doi.org/10.1016/j.atmosres.2022.106016.
Yoo, C., N. C. Johnson, C.-H. Chang, S. B. Feldstein, and Y.-H. Kim, 2018: Subseasonal prediction of wintertime East Asian temperature based on atmospheric teleconnections. J. Climate, 31, 9351–9366, https://doi.org/10.1175/JCLI-D-17-0811.1.
Zhang, D., and W. Song, 2018: Northern Hemisphere atmospheric circulation characteristics in 2017/2018 winter and its impact on weather and climate in China (in Chinese). Meteor. Mon., 44, 969–976, https://doi.org/10.7519/j.issn.1000-0526.2018.07.013.
Zhang, R., R. Zhang, and Z. Zuo, 2016: An overview of wintertime snow cover characteristics over China and the impact of Eurasian snow cover on Chinese climate (in Chinese). J. Appl. Meteor. Sci., 27, 513–526, https://doi.org/10.11898/1001-7313.20160501.
Zhang, Y., Z. Yin, H. Wang, and S. He, 2021: 2020/21 record-breaking cold waves in east of China enhanced by the ‘Warm Arctic-Cold Siberia’ pattern. Environ. Res. Lett., 16, 094040, https://doi.org/10.1088/1748-9326/ac1f46.
Zhou, B., and H. Wang, 2008: Interdecadal change in the connection between Hadley circulation and winter temperature in East Asia. Adv. Atmos. Sci., 25, 24–30, https://doi.org/10.1007/s00376-008-0024-6.
Zhou, B., Z. Wang, Y. Shi, Y. Xu, and Z. Han, 2018: Historical and future changes of snowfall events in China under a warming background. J. Climate, 31, 5873–5889, https://doi.org/10.1175/JCLI-D-17-0428.1.
Zhou, B., Z. Wang, B. Sun, and X. Hao, 2021: Decadal change of heavy snowfall over northern China in the mid-1990s and associated background circulations. J. Climate, 34, 825–837, https://doi.org/10.1175/JCLI-D-19-0815.1.
Zhu, Y., G. Tan, and Y. Wang, 2007: Variation of spatial mode for winter temperature in China and its relationship with the large scale atmospheric circulation (in Chinese). Adv. Climate Change Res., 3, 266–270.
Zuo, J., H. Ren, and W. Li, 2015: Contrasting impacts of the Arctic Oscillation on surface air temperature anomalies in southern China between early and middle-to-late winter. J. Climate, 28, 4015–4026, https://doi.org/10.1175/JCLI-D-14-00687.1.
Zuo, J., H.-L. Ren, B. Wu, and W. Li, 2016: Predictability of winter temperature in China from previous autumn Arctic sea ice. Climate Dyn., 47, 2331–2343, https://doi.org/10.1007/s00382-015-2966-6.